Skip to content

Code for Diffusion Action Segmentation (ICCV 2023)

License

Notifications You must be signed in to change notification settings

Finspire13/DiffAct

Repository files navigation

Code for Diffusion Action Segmentation (ICCV 2023).

Setup

  • Recommended Environment: Python 3.9.2, Cuda 11.4, PyTorch 1.10.0
  • Install dependencies: pip3 install -r requirements.txt

Data

  • Download features of 50salads, GTEA and Breakfast provided by MS-TCN and ASFormer: [Link1] [Link2]
  • Unzip the data, rename it to "datasets" and put into the current directory
DiffAct/
├── datasets
│   ├── 50salads
│   │   ├── features
│   │   ├── groundTruth
│   │   ├── mapping.txt
│   │   └── splits
│   ├── breakfast
│   │   ├── features
│   │   ├── groundTruth
│   │   ├── mapping.txt
│   │   └── splits
│   └── gtea
│       ├── features
│       ├── groundTruth
│       ├── mapping.txt
│       └── splits
├── main.py
├── model.py
└── ...

Run

  • Generate config files by python3 default_configs.py
  • Simply run python3 main.py --config configs/some_config.json --device gpu_id
  • Trained models and logs will be saved in the result folder

Trained Models

  • We provide some trained models in the trained_models folder

Citation

@inproceedings{liu2023diffusion,
  title={Diffusion Action Segmentation},
  author={Liu, Daochang and Li, Qiyue and Dinh, Anh-Dung and Jiang, Tingting and Shah, Mubarak and Xu, Chang},
  booktitle={International Conference on Computer Vision (ICCV)},
  year={2023}
}

License

MIT

About

Code for Diffusion Action Segmentation (ICCV 2023)

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages